Tasks in Sift are live entities, with space for collaboration from a range of stakeholders. They are created to track issues with assets where follow-up actions are required. They have space for in-depth observations including interactive charts, tables, and supporting documentation, as well as tracking which assets are affected, what the opportunity for revenue gain/avoided loss is, and intermediate actions along with their assignees, due date and states. A comment thread tracks recent observations and questions from all members of the team.

All of this information is essential for decision-making and delivering value, however it does not provide the bird’s eye view required for the roles with oversight and co-ordination responsibilities, such as the Asset Manager and the Operations & Maintenance Manager. This is where Sift AI Summaries come in. With Sift AI Summaries, we support these roles by maintaining an accurate overview of multiple and sometimes interacting issues affecting their projects.

Take a relatively simple case of a suspected main bearing failure. The workflow towards resolution involves several steps, starting with the detection of a potential fault condition through data analysis. A task report is created which instantiates multiple steps of follow-up work including initial trouble shooting, bearing inspection, replacement planning and procurement, execution which may include some civil works, continued monitoring to verify resolution, estimation of losses and cost summaries, lessons learned for the future, and more.

A typical issue on a production wind farm will contain much more detail than listed in this example – increasing the requirement for easily-digestible summaries.

As soon as the task is created, our AI model generates a Summary for any future viewers. Our AI uses the native multi-modal capabilities of the latest generation of LLMs, which means that the model can “see” the charts and other supporting information provided in the task. In addition, the LLM is provided access to historic tasks and other documentation, which it can reference to provide additional context to the reader.

Users can tailor their requests to Bitbloom’s AI to shape AI Summaries in a specific format, length, or placing particular emphasis on certain aspects, further supporting the Asset Managers and other Sift users to gain the specific insights they need.

Any updates to the task cause the Summary to be regenerated. Asset Managers and Operations & Maintenance Managers using Sift can therefore be sure they are not missing the latest updates. Analysts using Sift, meanwhile, get an easy-to-digest overview of the progress so far and can drop in their update and continue with their next task, leaving Bitbloom’s AI to summarise their contributions, allowing them to maintain their flow and increasing efficiency all round.